AI Extraction
Turn meeting notes, Slack threads, and documents into structured Jira tasks. AI reads your text and extracts actionable items with proper hierarchy.
SuperTemplates uses AI to read unstructured text — meeting notes, specs, Slack threads, emails — and extract structured Jira tasks with proper hierarchy, assignees, and details.
Indentation is the simplest way to tell the AI how your tasks relate to each other. Top-level lines become Epics, indented lines become Tasks, and further indented lines become Subtasks.
Top level → Epic
Lines with no indentation are treated as Epics — the highest-level initiative. Include the owner, approver, and target date. Example: Pawel is owning the API redesign initiative, targeting June 26th. Sarah approving.
One indent → Task (Story)
Indented lines under an Epic become Tasks or Stories. These are the user-facing goals that roll up to the initiative. Example: Users need faster response times on the dashboard endpoints - Marcus leading this, high priority, due May 8.
Two indents → Subtask
Further indented lines become Subtasks under the parent Task. Include assignee, priority, due date, sprint, and type (bug vs. task). Example: Implement response caching layer for frequently accessed data - Marcus, medium priority, due Apr 27, Initial sprint.
Pro tip: The AI reads natural language — you don't need a rigid format. Just keep your indentation consistent: same level = same hierarchy.
For maximum control, use numbered outlines. The numbering makes the hierarchy explicit — 1. for Epics, 1.1. for Tasks, 1.1.1. for Subtasks. This removes any ambiguity for the AI.
Number your Epics: 1., 2., 3.
Each top-level number becomes an Epic. Example: 1. Pawel is owning the API redesign initiative, targeting June 26th. Sarah approving. 2. Ewelina is leading the notifications overhaul, due July 10th. David approving.
Number your Tasks: 1.1., 1.2., 2.1.
Second-level numbers become Tasks nested under their parent Epic. Example: 1.1. Users need faster response times on the dashboard endpoints - Marcus leading this, high priority, due May 8.
Number your Subtasks: 1.1.1., 1.1.2.
Third-level numbers become Subtasks. Add metadata on the next line or inline. Example: 1.1.1. Implement response caching layer for frequently accessed data. Marcus | Medium | Due: Apr 27 | Sprint: Initial sprint - schemas api
Pro tip: You can mix numbering with natural language details. The AI picks up assignees, priorities, dates, sprints, and bug/task types from the surrounding text.
The AI extracts more than just task titles. It reads your text for assignees, priorities, due dates, sprints, labels, and issue types — all from natural language.
Assignees & approvers
Mentions like "Marcus leading this", "Emma taking this", or "Sarah approving" are matched against your Jira project members automatically.
Priorities
Language like "high priority", "needs fixing urgently", or "highest priority" maps to Jira priority levels (Lowest → Highest).
Due dates & sprints
"due May 8", "targeting June 26th", or "due Apr 27" become Jira due dates. Sprint names like "Initial sprint - schemas api" or "Design stuff sprint" are assigned automatically.
Issue types: bugs vs. tasks
The AI distinguishes bugs from tasks based on context. "Memory leak in the connection pool" or "Broken syntax highlighting" are flagged as bugs. Feature work stays as Tasks or Stories.
Pro tip: The more detail you include in your text, the less editing you'll need after extraction. Even rough notes produce good results — the AI fills in what it can infer.
AI extraction is powerful but not perfect. Results will vary depending on the AI model selected, the clarity and structure of your input notes, the complexity of the tasks described, and your Jira project configuration (available fields, issue types, sprints, and custom workflows). Some fields — such as sprint assignments, labels, or priorities — may not always be extracted, especially from loosely structured text. Always review and verify extracted tasks before creating them in Jira.
Pro tip: Enable the 'I will review and verify all extracted tasks' checkbox for an extra safety step before anything is created in Jira.